OpenAI Disclosure Saga: CISO Reportedly Sought to Sue Researcher, Later Apologized

Security researcher S1r1u5 (team Hacktron) verified a flaw via Codex on cloud that granted access to OpenAI's internal monorepo, claiming the ability to laterally move toward model weights, datasets and training recipes, but deliberately downloaded nothing sensitive and proved access only with a harmless PR before responsible disclosure. By his account, OpenAI's disclosure process was a nightmare requiring lawyers and journalists; the $6,500 bounty was never the issue. OpenAI's CISO (ex-Palantir) was reported to have favored suing the researcher, drawing community mockery, but per LiveOverflow has since reached out, apologized, and reconciled.

Confirmed

• The researcher verified repo access, proved it with a harmless PR, stopped, and followed responsible disclosure (m5)
• The researcher described OpenAI's disclosure process as extremely difficult, requiring lawyers and journalists; bounty was $6,500 (m3)
• LiveOverflow said the CISO proactively apologized and reconciled, noting his earlier critical thread was personal opinion (m6, m8)
• Hacktron founder S1r1u5 clarified Claude Code did not autonomously hack OpenAI; the key was human researcher rootxharsh's prior experience (m14)
• White-hat hacker s1r1us clarified that finding a flaw doesn't mean knowing how to protect such a complex organization; OpenAI has a more experienced security team (m12)

Unconfirmed

• Whether the CISO actually advocated suing the researcher rests mainly on reposted accounts, with no primary source
• No formal OpenAI commitment to fix the disclosure process appears in the material

Why it matters

• Veteran Robert Graham criticized OpenAI and Anthropic for abandoning the security community's 40-year principles of encouraging open disclosure and opposing retaliation (m2); researcher IceSolst argued OpenAI burned a rare trust-building opportunity and the CISO should be replaced (m11)
• A methodology debate ensued: Tal BeerySec argued 0-days call for responsible disclosure, not lateral movement, calling the PR on OpenAI's GitHub repo over the line (m13); jachiam0 (ex-OpenAI red team, now Anthropic) defended responsible disclosure even beyond bounty scope, called attacks on the CISO a distortion, and noted he had already warned about Slack lateral-movement risks (m7, m9, m10)
• Joshua Saxe mocked OpenAI's security maturity as resembling a company scaling too fast (m4), highlighting systemic tension between frontier AI labs and the external research community

2026-09-19 ~ 2026-09-21 · 14 related posts

Full story(4 episodes)→

• Episode 1: Three Hacktron Researchers Breach OpenAI's Internal Systems Using Claude Opus 5(2026-09-18, 59 posts)
• Episode 2: Hacktron AI Chained Two Bugs to Take Over OpenAI Accounts in 72 Hours; $6,500 Bounty Sparks Outcry(2026-09-18, 19 posts)
• Episode 3: Nathan Lambert: Closed-Source Models Are the True Tip of the AI Risk Iceberg(2026-09-18, 6 posts)
• Episode 4: OpenAI Disclosure Saga: CISO Reportedly Sought to Sue Researcher, Later Apologized(2026-09-19, 14 posts)

• LiveOverflow threads critique of how OpenAI's CISO fumbled the situation — xeophon · 2026-09-19
• [source] Security researchers blast OpenAI's "nightmarish" vulnerability disclosure process — evilsocket · 2026-09-19
• Security researchers slam OpenAI for burning trust over unreported vuln chain outreach — evilsocket · 2026-09-20
• Hacktron-OpenAI Security Drama: Insider Says OpenAI's Security Maturity Is Blitzscale-Level — joshua_saxe · 2026-09-20
• OpenAI safety researcher slams CISO pile-on over HacktronAI breach as 'total distortion' — jachiam0 · 2026-09-20
• jachiam0 doubles down: Slack logs at every major AI lab enable lateral movement, signed S3 URLs included — jachiam0 · 2026-09-20
• LiveOverflow says OpenAI CISO apologized over vulnerability disclosure fumble — xeophon · 2026-09-20
• [source] Hacktron Founder: Claude Code Didn't Hack OpenAI Alone — Human Zero-Day Intuition Was Key — gowthami_s · 2026-09-20
• OpenAI CISO mocked for reportedly wanting to 'threaten to sue' security researchers — kevinnbass · 2026-09-20
• White hat who breached OpenAI says frontier labs' security is alarmingly weak — Miles_Brundage · 2026-09-20
• Security veterans slam OpenAI and Anthropic for threatening researchers and discarding 40 years of disclosure norms — AlexTensor · 2026-09-20
• OpenAI vulnerability disclosure debate: researcher defends hack despite scope limits — jachiam0 · 2026-09-21
• Security Row: Was the 0-Day PR in OpenAI's GitHub Repo Responsible Disclosure or a Bridge Too Far? — joshua_saxe · 2026-09-21
• [source] OpenAI bug disclosure sparks CFAA debate as researcher defends good-faith repo access test — ns123abc · 2026-09-21